Potential field-based ship navigation early warning methods, equipment, media, and products

Abstract

This invention provides a ship navigation early warning method, device, medium, and product based on a potential field. The method includes: acquiring ship status, a set of static obstacles, a set of navigation rules, and information on environmental disturbances and uncertainties; performing dimensionless scaling and risk mapping normalization on the information; obtaining the static environmental potential field of the ship at the current trajectory point based on the difference between the shortest distance from the trajectory point to the set of static obstacles and the safety margin; constructing a dynamic interactive potential field based on the target ship's status and the urgency of its relative motion; triggering navigation rules and generating rule constraints based on the encounter state identification results to construct a rule-constrained potential field; expanding and correcting the safety margin based on uncertainty information to construct an uncertainty-corrected potential field; fusing these elements to obtain the total potential field; and obtaining a risk index and early warning level based on the total potential field. This technical solution achieves risk early warning based on a computable multidimensional safety potential field, improving compliance, stability, and robustness in complex waters and under multi-ship encounter conditions.

Description

Technical Field

[0001] This invention relates to the field of intelligent navigation and collision avoidance technology for ships, and in particular to a ship navigation early warning method, device, medium and product based on potential field. Background Technology

[0002] With the development of intelligent navigation, shore-based remote assisted driving, and unmanned autonomous navigation, intelligent ships need to make safe, compliant, and executable collision avoidance decisions in complex waters such as ports and narrow waterways, or in waters with heavy traffic. Existing related technologies mainly include: hazard assessment and rule-based strategies based on nearest point and nearest time, local planning based on artificial potential fields, path planning based on sampling / graph search, and constraint optimization control based on model predictive control (MPC).

[0003] (1) Risk assessment and rule-based strategies based on indicators such as DCPA / TCPA / CRI: This type of method usually calculates the closest point distance and the time to reach the closest point based on the relative motion of the two ships, and combines empirical thresholds or comprehensive risk indicators to classify risks and issue warnings, and then chooses to give way or maintain the course based on navigation rules. Its advantages are that the indicators are intuitive and easy to implement; however, it has obvious limitations in complex waters and multi-ship encounters: First, the indicators are mostly for the relationship between two ships. When facing multi-ship coupling, channel boundaries, static obstacles and environmental disturbances, it is difficult to form a unified risk cost and an executable trajectory; Second, rule triggering and conflict handling often rely on empirical criteria, making it difficult to form calculable constraints and be consistent with trajectory optimization; Third, the indicator thresholds are strongly dependent on the scenario and ship type, and lack portability and robustness.

[0004] (2) Artificial Potential Field (APF) and Risk Field Methods: This type of method generates local obstacle avoidance paths by constructing target attraction potential and obstacle repulsion potential under the guidance of potential field gradient. It has the advantages of good real-time performance and low computational cost, and is therefore widely used in robot obstacle avoidance and ship local collision avoidance. However, in ship navigation scenarios, common problems of potential field methods are more prominent: First, the dimensions of factors such as static obstacle geometry, channel boundary, target ship interaction urgency, and navigation rules are different. Simple weighted superposition leads to "reskinned" risk fields, which lack a unified physical / risk interpretation. Second, the superposition of multiple obstacles and multiple ships is prone to local minima and saddle points, resulting in stagnation, edge oscillation, or back-and-forth swinging. Third, when the potential field gradient is directly mapped to the control quantity, the ship's underactuated characteristics and maneuvering constraints such as minimum turning radius, rudder angle / rudder speed limit, and speed change limit are often ignored, resulting in unexecutable generated trajectories or deviations from the plan after execution.

[0005] (3) Sampling / Search Path Planning and MPC-type Constraint Optimization: Sampling and graph search can obtain globally feasible paths, and MPC can output executable controls under dynamic and manipulation constraints. Its shortcomings are: First, if reasonable risk cost modeling and rule constraint modeling are lacking, the obtained trajectory may meet geometric safety but have insufficient compliance or interpretability; Second, if multi-source risk factors are directly used as multi-objective / multi-constraint input optimizers, it is easy to encounter difficulties in parameter selection, increased computational complexity, and insufficient real-time performance; Third, the lack of a unified interface for rule triggering, responsibility judgment, and constraint generation leads to a break in the "evaluation-planning-control" chain.

[0006] Furthermore, navigation in complex waters is characterized by significant uncertainties, including wind, waves, current disturbances, reduced visibility, sensor measurement errors, communication delays, and execution errors. This necessitates robust and conservatively controllable risk assessment and collision avoidance decisions. Existing methods often rely on experience to increase safety distances or make ex-post corrections, lacking a unified framework that incorporates uncertainties into the risk potential field and decision optimization in a calculable manner. Summary of the Invention

[0007] To address one or more problems in the prior art, embodiments of the present invention provide a ship navigation early warning method, device, medium, and product based on a potential field.

[0008] To achieve the above objectives, on the one hand, a ship navigation early warning method based on a potential field is provided, including:

[0009] S1, acquire the ship's status, target ship status, set of static obstacles, set of navigation rules, and environmental disturbance and uncertainty information; the ship's status and target ship status include the ship's ground-fixed coordinates, heading, and speed; wherein, static obstacles include one or more of the following: channel boundaries, shorelines, shoals, reefs, bridge piers, and harbor structures; the environmental disturbance and uncertainty information includes a set of estimated covariances or standard deviations obtained by statistically analyzing wind, wave, and current estimations, visibility, sensing errors, and / or execution errors;

[0010] S2, using the selected length scale Speed ​​scale With time scale The ship's state, target ship state, static obstacle set, navigation rule set, and environmental disturbance and uncertainty information obtained in step S1 are processed to be dimensionless, resulting in dimensionless state quantities and dimensionless indices derived from these state quantities. The dimensionless indices include: dimensionless distance indices, dimensionless time indices, and dimensionless uncertainty scales. These dimensionless indices are then input into a risk mapping function. Perform normalization;

[0011] Wherein, the risk mapping function for any dimensionless index s for:

[0012] ;

[0013] in, , The gain coefficient is greater than 0; This represents a nonlinear saturation function or activation function;

[0014] S3, based on trajectory points Shortest distance to static obstacles in the static obstacle set and preset safety margin The difference is used to determine the trajectory points using the risk mapping function. static risks And based on the static risk and the predetermined logarithmic potential energy transformation formula, the ship's trajectory point is obtained. Static environment potential field The predetermined logarithmic potential energy transformation formula is as follows:

[0015] ;

[0016] Among them, the The dimensionless shortest distance obtained under the length scale L determined in step S2, wherein The dimensionless safety margin obtained under the length scale L;

[0017] in, ;

[0018] ;

[0019] in, For a predetermined minimum positive number, For the static environmental potential field of index s,

[0020] For the shortest distance The static potential field of the environment;

[0021] S4, based on the urgency index of the relative motion between the two ships and the preset interaction safety distance, obtains the interaction risk between the ship and each target ship j. According to interaction risk The dynamic interaction potential field between the ship and each target ship j is obtained using the predetermined logarithmic potential energy transformation formula. The dynamic interactive potential field between the ship and all target ships is obtained through a predetermined multi-objective weighted fusion method. The urgency index includes the dimensionless nearest point distance between the two ships obtained using the length scale L. and the time of closest arrival of the dimensionless point ;

[0022] S5, based on this ship and each target ship The encounter status recognition result triggers navigation rules, generating a target ship. rule constraint set ; and based on control quantity and the set of rules and constraints Constructing the permissible maneuver set and prohibited motor assembly To then construct the target ship Corresponding distance-based violation Then according to The target ship is obtained by using the predetermined logarithmic potential energy transformation formula. The regular potential field components The rule constraint potential field of this ship for all target ships is obtained by weighted fusion of the rule potential field components of all target ships. The control quantity This includes yaw rate and longitudinal acceleration. Among them:

[0023] ;

[0024] ;

[0025] in, This is the penalty coefficient; Indicates the target ship The corresponding urgency weight, The total number of target ships; express violation The depth, where, when hour, ;otherwise, ; Indicates control quantity With permitted maneuver sets The minimum Euclidean distance is defined as:

[0026] in express Compared to the prohibited maneuver assembly Signed distance to the boundary: when When the value is positive and equal to The minimum distance to the boundary of the forbidden set; for any element in; when When the value is 0 or negative, it is used to make Output 0 if no violation occurs; Indicates the target ship The urgency weight, used to reflect the strength of the contribution of different target ships to the rule constraints, can be obtained by normalizing the interaction risk or urgency index. Furthermore, when multiple target ships are in the same encounter state and trigger the same rule constraint, their... , They can be the same; otherwise , Different, therefore and different; It is a Euclidean norm;

[0027] S6, based on the aforementioned environmental disturbance and uncertainty information, for distance-related indicators... Perform a conservative correction to obtain the corrected effective distance. and use the aforementioned in distance-based risk mapping Replace the distance-type indicator To obtain an uncertainty-corrected potential field ;in:

[0028]

[0029] in, The distance-type indicators are those obtained after dimensionless processing in step S2, including the shortest distance to static obstacles. , and the target ship nearest point distance and channel boundary margin At least one of them; For the distance-type index The standard deviation is obtained from the state estimation covariance propagation; when When static obstacles are involved, the state estimation covariance shall include at least the ship's state covariance. ;when Involving the target ship At that time, the state estimation covariance includes at least the ship's state covariance. Covariance with the target ship's state and form relative covariance ; is the confidence coefficient.

[0030] S7, the static environment potential field, the dynamic interaction potential field, the rule constraint potential field, and the uncertainty correction potential field are weighted and fused according to predetermined weights to obtain the total potential field. ;

[0031] S8 represents the total potential field at time t. Mapping to a risk index via exponential saturation or the Sigmoid function. , And by comparing risk indices The warning level is output based on the value of the preset risk index threshold.

[0032] Preferably, in the potential field-based ship navigation early warning method, step S8 further includes one or more of the following steps:

[0033] Multiple risk index thresholds can be preset to output multiple warning levels; the higher the risk index threshold, the higher the warning level.

[0034] Output the percentage of each potential field's contribution to the total duration;

[0035] Using the potential field evolution as a risk index To provide early warning, at this time:

[0036] ;

[0037] The total potential field at time t ; , and These are the preset weighting coefficients.

[0038] Preferably, the ship navigation early warning method based on potential field further includes, after step S8:

[0039] S9, when the risk index Greater than or equal to the preset collision avoidance trigger risk threshold or the rate of change of the risk index Greater than or equal to the preset collision avoidance trigger risk change rate threshold Collision avoidance optimization is triggered in a timely manner, and a multi-objective collision avoidance optimization problem is solved under preset ship kinematics or dynamics constraints, maneuvering constraints, and rule constraints to generate executable heading, speed, rudder angle, and / or yaw rate control commands, which are then output to the ship control system; wherein, the risk index change rate is defined as:

[0040] ;

[0041] To control the period or time step; the multi-objective collision avoidance optimization problem uses a control sequence. Let be the decision variables, where , Yaw angular velocity, For longitudinal acceleration, the prediction step size is... The predicted trajectory point position vector is , The reference point is the centerline of the navigation channel or the target trajectory; the objective function used in the multi-target collision avoidance optimization is:

[0042] ;

[0043] in, For the first The total potential value at the step prediction time. These are the weighting coefficients for the corresponding items. Let be the Euclidean norm, and let... , The control quantity is output and executed in the previous control cycle, so that the control increment term has a computable meaning at time k=0.

[0044] Preferably, in the potential field-based ship navigation early warning method, step S4 includes:

[0045] ;

[0046] ;

[0047] in, This is the distance between the closest point between the current vessel and the target vessel j; To estimate the time of closest approach between this vessel and the target vessel j;

[0048] Define the time window function:

[0049] ;

[0050] Preset interaction safety distance for:

[0051] ;

[0052] ， ;

[0053] ;

[0054] ;

[0055] ;

[0056] ;

[0057] ;

[0058] in, and These are the main vessel and the target vessel, respectively. Length, For the ship's speed, For the target ship speed, Indicates the target ship The relative position vector of the ship itself. Indicates the target ship The relative velocity vector with respect to the ship itself; This is the time constant for the ship's reaction or execution; Indicates the relationship between this vessel and the target vessel The dimensionless uncertainty scale of the relative position is obtained by propagating the covariance of the position estimates of the ship and the target ship, and is used to make a conservative correction to the interaction safety distance. , and These are preset weighting coefficients; For indicator functions, only when the following conditions are met The value of the indicator function is only non-zero when the specified time is reached; This is the upper limit of the time window; The decay time constant; This is a preset constant; The total number of target ships; and This is the preset gain coefficient; The preset time threshold; where, All are dimensionless indices obtained after dimensionless processing in step S2, and and Expressed on the same length scale.

[0059] Preferably, in the aforementioned ship navigation early warning method based on a potential field, step S6 involves obtaining an uncertainty-corrected potential field by constructing an upper bound term for a risk index. ;in:

[0060] ;

[0061] in, The penalty weight for uncertainty has a value ≥ 0; is the trace of the covariance matrix.

[0062] Preferably, in the potential field-based ship navigation early warning method, in step S3, if a channel boundary exists, then the warning is based on signed distance. Constructing the boundary potential field To strengthen penalties for crossing the boundary, and Merging Among them, when the trajectory point When within the boundary, When on the boundary, When crossing the boundary, .

[0063] Preferably, the ship navigation early warning method based on potential field further includes:

[0064] The multi-objective collision avoidance optimization problem is solved using SQP, iLQR, gradient projection, or rolling MPC methods to obtain the control sequence. Only the first item is executed. And will be updated on a rolling basis in the next cycle;

[0065] When detected Furthermore, if the risk does not decrease or the control quantity frequently changes direction, tangential guidance, damping term, random disturbance restart, or replanning strategy will be triggered, and the generated guidance quantity will be projected onto the feasible control set. ;

[0066] in, Represents the total potential field gradient, express The Euclidean norm; It is a pre-defined small positive number.

[0067] On the other hand, an electronic device is also provided, including a memory and a processor, the memory storing at least one program, the at least one program being executed by the processor to implement the steps of the potential field-based ship navigation early warning method as described above.

[0068] In another aspect, a computer-readable storage medium is also provided, wherein at least one program is stored therein, the at least one program being executed by a processor to implement the steps of the potential field-based ship navigation early warning method as described above.

[0069] In another aspect, a computer program product is also provided, comprising a computer program, characterized in that, when the computer program is executed by a processor, it implements the steps of the potential field-based ship navigation early warning method as described above.

[0070] The above technical solution has the following technical effects:

[0071] The technical solution of this invention achieves the integration of static geometric redundancy, dynamic urgency, rule constraints, and uncertainty within a unified risk cost space through dimensionless transformation and risk mapping normalization. This avoids the inexplicability and non-transferability caused by splicing heterogeneous experience. It can unify static environment, dynamic interaction, navigation rule constraints, and uncertainty factors into a computable, interpretable, and executable multidimensional safety potential field, and realize a closed-loop integrated method and system for risk warning and collision avoidance decision-making. This improves compliance, stability, and robustness in complex waters and multi-vehicle encounters.

[0072] Further embodiments of this invention utilize a "meeting state recognition—rule triggering—constraint generation—potential field penalty / hard constraint" chain to make compliance a computable object and enter an optimization closed loop, reducing the risk of "safe but non-compliant"; by explicitly introducing ship maneuvering and dynamic constraints in the optimization process, the executable output trajectory or maneuvering command is ensured, avoiding non-executability caused by direct control of the potential field gradient; through local optima and oscillation detection, as well as tangential guidance / damping / restart mechanisms, stability is improved in scenarios involving narrow waterways, multiple obstacles, and dense interactions between multiple ships; through uncertainty expansion and risk upper bound mechanisms, robustness and conservative controllability are improved under conditions of wind, wave, and current disturbances, perception errors, and execution errors; and through risk contribution decomposition output, risk source ranking and compliance explanation information are provided, enhancing interpretability and auditability. Attached Figure Description

[0073] Figure 1 This is a flowchart illustrating a potential field-based ship navigation early warning method according to an embodiment of the present invention.

[0074] Figure 2 This is a schematic diagram of the structure of a system for implementing a ship navigation early warning method according to an embodiment of the present invention;

[0075] Figure 3 This is a schematic diagram of the multidimensional safety potential field decomposition and total potential field fusion in a potential field-based ship navigation early warning method according to an embodiment of the present invention.

[0076] Figure 4 This is a schematic diagram of the encounter situation identification, rule triggering and constraint generation process in a potential field-based ship navigation early warning method according to an embodiment of the present invention.

[0077] Figure 5 This is a schematic diagram illustrating the risk index and warning level determination in a potential field-based ship navigation early warning method according to an embodiment of the present invention.

[0078] Figure 6 This is a schematic diagram of the rolling optimized collision avoidance closed-loop structure in a potential field-based ship navigation early warning method according to an embodiment of the present invention. Detailed Implementation

[0079] To further illustrate the various embodiments, the present invention provides accompanying drawings. These drawings are part of the disclosure of the present invention, primarily used to illustrate the embodiments and to explain the operating principles of the embodiments in conjunction with the relevant descriptions in the specification. With reference to these drawings, those skilled in the art should be able to understand other possible implementations and the advantages of the present invention. Components in the drawings are not drawn to scale, and similar component symbols are generally used to represent similar components.

[0080] The present invention will now be further described in conjunction with the accompanying drawings and specific embodiments.

[0081] Example 1:

[0082] Figure 1 This is a flowchart illustrating a potential field-based ship navigation early warning method according to an embodiment of the present invention. Figure 1 The method in this embodiment controls the cycle. The rolling execution includes the following steps:

[0083] S101, Data Acquisition and Unified Representation;

[0084] Obtain and uniformly represent the following inputs:

[0085] 1) Ship's condition and manipulation constraint parameters; for example: , , , wait; These are the ship's transverse and longitudinal coordinates, heading, and speed, respectively; in one specific implementation, These are the x and y coordinates in Earth-fixed coordinate system.

[0086] Earth-fixed coordinate system: (East-North Plane);

[0087] 2) Target ship assembly ; Let the state of target ship j at time t be: ; The number of target ships;

[0088] 3) Geometric set of static obstacles such as channel boundaries, shorelines, shoals, and bridge piers. and channel centerline / recommended route ;

[0089] 4) Rule set One implementation includes machine-readable rule entries from the International Regulations for Preventing Collisions at Sea (ICMLS), port regulations, or rules concerning traffic constraints, such as the lane separation scheme.

[0090] 5) Environmental disturbances and uncertainties In one specific implementation, it includes wind, wave, and current estimation, visibility, and statistics on sensing and execution errors, and outputs the state estimation covariance. Or the set of standard deviations.

[0091] 6) Output: Unified State Library ; This includes all of the above.

[0092] S102, coordinate transformation, dimensionless transformation and risk mapping normalization;

[0093] Transform the Earth-fixed coordinates to the ship's coordinate system. To obtain the relative position With relative velocity Select scale: length ,speed ,time And the relative position and relative velocity obtained in step S102 are dimensionless:

[0094] in, For the target ship dimensionless components of relative position Let the relative velocity be a dimensionless vector. This is the minimum speed of the vessel; in one specific implementation, the above length is used. ,speed and time The corresponding dimensionless scaling scales L, V, and T are used.

[0095] Ship coordinate system Among them, the bow is Positive direction, starboard side positive direction;

[0096] For any physical indicator s, such as distance-based, time-based, boundary-crossing-based, and violation-based indicators, risk mapping is applied respectively. Furthermore, it can be transformed into potential energy using logarithmic potential transformation, where the risk mapping function is defined as:

[0097] ;

[0098] ;

[0099] in As a safety threshold, in one specific implementation, it can adapt to different scenarios, ship types, and / or rules; This represents the steepness coefficient. The smaller the distance, the greater the risk; if it represents the "degree of violation," then use the opposite sign or rewrite it. To map risks The potential field function is converted into potential energy.

[0100] Output: Dimensionless feature set after risk normalization .

[0101] In this embodiment of the invention, a dimensionless scale unification is employed to ensure that multi-source heterogeneous indicators such as distance, time, velocity, uncertainty scale, and rule thresholds can be risk-mapped and potential-field fused within the same numerical domain. Specifically, a length scale is selected. Speed ​​scale Time scale The dimensionless quantities are obtained by dividing position, distance, speed, time, and their derived indices by their corresponding scales. Angular quantities, such as heading angle and relative azimuth angle, are inherently dimensionless and can retain their original values. In this embodiment, the dimensionless quantities are used for subsequent static shortest distance calculation, dynamic CPA urgency calculation, risk mapping, potential field construction and fusion, risk warning, and collision avoidance optimization cost and constraint calculation. For simplicity, unless otherwise stated, the symbols for length, speed, time, and their derived indices in the following formulas refer to the dimensionless quantities after dimensionless processing, and are no longer marked with "~". Angular quantities, such as heading angle and relative azimuth angle, are inherently dimensionless and can retain their original values. Before outputting control commands, the dimensionless control quantities are de-dimensionalized according to their corresponding scales to obtain the actual executable control commands.

[0102] Specifically, for ease of description, the dimensionless symbol “” can be omitted in subsequent steps S103 to S111. However, the distance, speed, and time parameters involved are all calculated and used in dimensionless form under the scale of step S102, specifically including:

[0103] (1) In step S103, the shortest distance of the static obstacle Channel boundary redundancy and safety margin All according to length scale Expressed as a dimensionless quantity and used as a risk mapping function Input;

[0104] (2) In step S104, the distance to the nearest point and safe distance for interaction All according to length scale Expressed as a dimensionless quantity, time to reach the nearest point and time window parameters All according to time scale It is expressed as a dimensionless quantity, and the interaction risk is calculated accordingly. With dynamic potential field ;

[0105] (3) In steps S107 to S109, the total potential field Risk Index Calculations are performed in a dimensionless domain.

[0106] (4) In steps S110 to S111, the potential cost in the collision avoidance optimization objective function Dimensionless variables are used for calculation; before outputting control commands, the dimensionless control quantities are dedimensionalized according to the corresponding scale to obtain the actual executable control commands.

[0107] S103, Constructing a static environmental potential field ;

[0108] The static environment includes shoreline, shoals, reefs, bridge piers, harbor basin structures, and channel boundaries.

[0109] For any point Or predict trajectory points Calculate to the obstacle set Shortest distance:

[0110]

[0111] Construction safety margin In one specific implementation, the safety margin is at least related to the ship's length, speed, reaction time, and uncertainty; in another specific implementation, a safety margin is pre-set, and during the process, the safety margin adaptively adjusts with speed and / or ship type.

[0112] ;

[0113] in, , and These are the corresponding weighting coefficients; This refers to the length of the ship, i.e., its length. The reaction or execution time constant, The reaction / execution time constant. For the scale of location uncertainty;

[0114] Obtain the static risk and potential field:

[0115] ;

[0116] ;

[0117] in Set the threshold according to "small distance → high risk". ;

[0118] If a channel boundary exists, it is based on signed distance. Constructing the boundary potential field Regarding crossing the boundary ( Intensify punishment and Merging .

[0119] In one specific implementation, the boundary risk mapping can be defined as:

[0120]

[0121] in, Steepness coefficient, For the boundary safety margin (dimensionless). The out-of-bounds penalty coefficient; when Time indicates exceeding the boundary. Boundary potential field weight. Used to control the boundary potential field in The proportion of.

[0122] ;

[0123] In one specific implementation:

[0124] .

[0125] Among them, the shortest distance to the obstacle or boundary: The channel boundary has a marked distance: Its value is positive within the boundary and negative when it exceeds the boundary;

[0126] Output: .

[0127] in, Figure 4 This diagram illustrates the process of encounter situation identification, rule triggering, and constraint generation between the vessel and the target vessel. Figure 4 The process includes: 301, calculation of encounter elements; 302, identification of encounter types, including: face-to-face, intersection, or overtaking; 303, retrieval and triggering of rule entries; 304, determination of yielding responsibility or priority, including conflict resolution; 305, generation of permitted maneuver sets. ;306, Generate a set of prohibited maneuvers. 307. Generate constraint set C, including soft constraint penalties and / or hard constraint boundaries; 308. Output the generated constraint set to the potential field or optimization module, and then output the corresponding constraint set to the potential field construction and fusion module 40 for potential field construction and fusion and collision avoidance decision optimization module 60, so as to provide subsequent collision avoidance decision optimization.

[0128] Specifically, including:

[0129] S104, Constructing a dynamic interactive potential field ;

[0130] For each target ship Calculate the urgency index of the relative motion between the two ships, including but not limited to the distance to closest point (DCPA) and the time to reach the closest point (TCPA) (when the relative speed is non-zero). This indicates the TCPA of this ship relative to the target ship j; Indicates the DCPA of this vessel relative to the target vessel j:

[0131] ;

[0132] ;

[0133] Relative quantity: , ;

[0134] Furthermore, a prediction window is set. With time window function, i.e. decay function To emphasize future risks, The attenuation constant is:

[0135] ;

[0136] This is an indicator function, which focuses only on functions that are approaching and will appear in the future. The target that reaches the closest point within a given time.

[0137] Constructing interactive safe distance In one specific implementation, the safe distance for interaction The interaction risk of the target ship is at least related to the ship's dimensions, reaction time, and uncertainties. .

[0138] ;

[0139] This formula reflects that the required safety distance is not fixed, but dynamically calculated, covering safety margins in several different dimensions. σ is the standard deviation, used to represent the scale of uncertainty in relative position; it is dimensionless. , and This is a weighting coefficient, a positive number, used to adjust the contribution of each item.

[0140] Dynamic risk mapping:

[0141] ;

[0142] ;

[0143] ;

[0144] To emphasize that only future urgent interactions within the prediction window are considered, and to apply a decaying weighting to time urgency, a time window weighting function is defined:

[0145] ;

[0146] in For indicator functions, The upper limit of the time window, is the time decay constant.

[0147] Dynamic potential field (logarithmic potential):

[0148] ;

[0149] Multi-objective fusion employs soft maximum or logarithmic and exponential functions to avoid linear superposition saturation:

[0150] ;

[0151] Where the coefficient Used to control the degree of "maximization"

[0152] Output: .

[0153] The above construction preserves the physical interpretability of DCPA / TCPA, but through Safe distance It is integrated with "soft-max" to make it stable and differentiable under multiple ship superposition and uncertainty, which makes it easy to enter the optimizer.

[0154] S105, Encounter Situation Recognition and Rule Triggering;

[0155] In the local coordinate system, based on the relative bearing of the target ship Relative heading difference The relative motion direction and traffic organization information will be used to identify the encounter type as meeting, crossing, overtaking, etc., and trigger the corresponding set of rule entries. Output the rule triggering result and the yield responsibility identifier, and generate a set of rule constraints. This includes at least: permitted / prohibited directions of movement, minimum yielding or holding conditions, and the results of handling rule conflict priorities.

[0156] S106, Construct a rule-constrained potential field ;

[0157] Based on this ship and each target ship The encounter status recognition result triggers navigation rules, generating a target ship. rule constraint set The set of rule constraints It should include at least: permitted / prohibited maneuvering directions, minimum yielding maneuvers, conditions for maintaining course and speed, and priority sets when multiple rules are triggered. Furthermore, based on control variables and the set of rule constraints Construct the target ship within the controlled space. Corresponding set of permitted maneuvers With prohibited motor assembly Based on this, distance-based violation is defined. ; wherein, the control quantity Including yaw rate and longitudinal acceleration, denoted as:

[0158]

[0159] The target ship The corresponding distance-based violation is defined as:

[0160]

[0161] in, This is the penalty coefficient; Indicates control quantity With permitted maneuver sets The minimum Euclidean distance is defined as:

[0162]

[0163] express Compared to the prohibited maneuver assembly Signed distance to the boundary: when When the value is positive and equal to The minimum distance to the boundary of the forbidden set; when When the value is 0 or negative, it is used to make Output 0 if there is no violation.

[0164] To map the violation degree to regular potential field components, the Sigmoid function described in step S102 is used. Constructing rule-based risk mapping:

[0165] ;

[0166] in, This is the gain coefficient. The violation reference threshold is used; and the target ship is obtained through a predetermined logarithmic potential energy transformation. Regular potential field components:

[0167] ;

[0168] in It is a predetermined minimum positive number.

[0169] By weighted and fused the regular potential field components of all target ships, the regular constraint potential field is obtained:

[0170] ;

[0171] in, The total number of target ships; For the target ship The urgency weight is used to reflect the strength of the contribution of different target ships to the rule constraints. It can be obtained by normalizing interactive risk or urgency indicators (e.g., by...) (obtained through normalization).

[0172] Furthermore, when multiple target ships are in the same encounter state and trigger the same rule constraint, their , They can be the same; otherwise , Different, therefore and The differences reflect the variations in rule constraints for different target vessels.

[0173] Output: .

[0174] S107, Constructing an uncertainty-corrected potential field ;

[0175] Based on environmental disturbances and uncertainty information, a conservative correction is made to the distance-based safety margin and risk assessment to obtain the uncertainty-corrected potential field. In one specific implementation, the uncertainty correction includes at least one of the following methods or a combination thereof:

[0176] (1) Conservative correction for uncertainty in distance-based safety margin:

[0177] For distance-related indicators The effective distance is obtained by applying uncertainty inflation correction. :

[0178]

[0179] in, These are distance-related metrics, including the shortest distance to static obstacles. nearest point distance Channel boundary redundancy At least one of them, Here is the confidence coefficient. Distance-related metrics standard deviation Obtained by propagation of state estimation covariance; the distance term in risk mapping or potential field construction is used. Replace with This is done to form a conservative potential field or an equivalent risk upper bound correction, thereby obtaining an uncertainty correction effect.

[0180] (2) Construction of the upper bound of the risk index:

[0181] In another specific implementation, the uncertainty correction potential field is obtained by constructing an upper bound term for the risk index:

[0182]

[0183] in, For the penalty weight of uncertainty, ; Covariance matrix The trace; when only static obstacles are considered. This is the covariance of the ship's state; when the target ship is involved... When the relative amount is, For this vessel and the target vessel The relative covariance, and under independent estimation conditions .

[0184] The uncertainty information includes: the covariance matrix of position, heading, and / or velocity estimates. and the standard deviation of the corresponding state variables. .

[0185] Output: .

[0186] S108, the total potential field is obtained by fusing the multi-dimensional potential fields;

[0187] like Figure 3 As shown, either weighted fusion or hierarchical fusion methods are used:

[0188] ;

[0189] In one implementation, the initial values ​​of each weight are preset; in another, they are adaptively adjusted according to the scenario, the urgency of the risk, and the rule triggering state; for example, they increase when the risk increases. Increase in narrow waterways In this embodiment, the total potential field further includes the target guiding potential field. And / or manipulate a comfortable and smooth potential field When there is no need to introduce target guidance or manipulation smoothing constraints, the corresponding weights can be set... and / or Alternatively, the corresponding potential field components may not participate in the fusion. Furthermore, in one embodiment, and It is only used for collision avoidance optimization modeling in step S9 to achieve smooth target tracking and manipulation, and does not participate in the risk index mapping in step S8; in another embodiment, and It can also be used to construct the overall potential field and for risk mapping.

[0190] Output: .

[0191] in, , , , , and These are the static environmental potential fields. 401. Dynamic Interactive Potential Field 402. Rule-constrained potential field 403. Uncertainty-corrected potential field 404. Target-Guiding Potential Field 405 and Comfort / Smooth Potential Field The weights of the corresponding potential field of 406, and the weight parameter set of 411. , , , , , The corresponding potential field and weights are input to the potential field fusion operator 407, and fused through weighted summation and / or hierarchical fusion to obtain the total potential field. 408; Output the contribution decomposition of each potential field based on its proportion. 412; Total potential field. The risk index 410 is mapped to the risk mapping operator 409.

[0192] Regarding the target guiding potential field Construction of 405:

[0193] Target-guided potential field Used to constrain ships to follow the centerline of the channel or a preset reference track. Navigation; for any point Or predict trajectory points To find its reference trajectory The nearest projection point on ,in:

[0194] ;

[0195] And define the lateral deviation Then the target-guided potential field can be constructed as follows:

[0196] ;

[0197] in This indicates the deviation of the ship's course from the tangential direction of the reference track. is the weighting coefficient; in embodiments where only lateral deviation needs to be constrained, it can be taken as . .

[0198] Regarding the comfort / smooth potential field Construction of 406:

[0199] Comfort / Smooth Potential Field Used to suppress excessively large or rapid manipulation changes; a control quantity is set. ,in Yaw angular velocity, It is longitudinal acceleration, and , To correspond to the upper limit, a normalized control quantity can be defined. and construct

[0200] ;

[0201] in , These are the weighting coefficients. To control the cycle; in embodiments where only the rate of change of manipulation needs to be suppressed, the following can be taken: .

[0202] S109, Risk Index Calculation and Early Warning Classification;

[0203] Will Mapped to risk index :

[0204] .

[0205] To characterize the evolution of the potential field, the rate of change of the total potential field (the evolution of the potential field) is defined as:

[0206]

[0207] Meanwhile, the rate of change of the risk index can be defined as:

[0208] The warning level is output based on a pre-set risk index threshold; in one specific implementation, the risk index threshold includes multiple thresholds, forming a risk index threshold set. The output warning levels include multiple levels, forming a warning level set. In this implementation, the higher the warning level, the greater the risk.

[0209] Warning level:

[0210] ;

[0211] in, Here, R represents the risk index value.

[0212] Explainable contribution from output:

[0213] ;

[0214] Output: Warning level, risk index, and contribution breakdown. Among them, It is the product of one of the potential fields in each dimension that generates the total potential field and its corresponding weight value when generating the total potential field; This represents the total potential field generated by weighting the potential fields in each dimension; It is a very small positive value.

[0215] in, Figure 5 A diagram illustrating the risk index and warning level determination is provided. Figure 5 The diagram shows the corresponding risk index curve R(t) 501; the threshold line for each risk index threshold 503; the warning level 504; and the collision avoidance trigger point. 505; and when the risk index value reaches At that time, execute step 506 to output an early warning.

[0216] In one specific implementation, regarding the risk index Evolutionary terms can also be added to enable early warning:

[0217] ;

[0218] S110, Collision Avoidance Triggering and Multi-Target Optimization Modeling;

[0219] when or Collision avoidance optimization is triggered on time; and The preset collision avoidance trigger threshold is defined; the prediction step size is the discrete-time prediction length. Control cycle, i.e., time step Using ship kinematics or dynamics models and maneuvering constraints, the following objective function is constructed:

[0220] ;

[0221] make , This refers to the control variables executed in the previous cycle.

[0222] The objective function includes: the total potential cost accumulated along the predicted trajectory, the cost of trajectory deviation or flight time, the cost of control energy consumption or control amplitude, and the cost of smoothing the rate of change of control.

[0223] in, Used as a reference point for the centerline of the waterway or the target track;

[0224] And apply one or more of the following constraints: rudder angle, rudder angle change rate, rudder speed, yaw rate, longitudinal acceleration, velocity boundary, minimum turning radius; where, if a hard constraint method is used, it also includes rule hard constraints.

[0225] In one specific implementation, a simplified kinematic or dynamic model that can be implemented is selected, such as:

[0226] ;

[0227] ;

[0228] ;

[0229] ;

[0230] k corresponds to the kth control cycle or step size.

[0231] Example of control constraints:

[0232] ;

[0233] ;

[0234] ;

[0235] The subscript max is used to indicate the maximum possible value; min is used to indicate the minimum possible value.

[0236] Output: Optimization problem .

[0237] S111, Solvency and Executability Output;

[0238] Solve using SQP, iLQR, gradient projection, or rolling MPC. Obtain the control sequence Only the first item is executed. And will be updated in the next cycle.

[0239] Output: Executable control commands; in one implementation, the executable control commands include heading, speed, rudder angle and / or yaw rate, and a sequence of trajectory points.

[0240] S112, Local Optimality / Oscillation Detection and Suppression;

[0241] If detected If the risk does not decrease or the control quantity frequently changes direction, then tangential guidance, damping term, random disturbance restart, or reprogramming strategy will be triggered; and the generated guidance quantity will be projected onto the feasible control set. To ensure executability.

[0242] In one specific implementation, a tangential guiding term is given:

[0243] ;

[0244] ;

[0245] in This is the operator for rotating 90 degrees.

[0246] The minimum value used in the formulas in this embodiment They do not need to be equal.

[0247] Final guidance:

[0248] ;

[0249] in, The comprehensive guiding vector is used to characterize the term guided by the gradient of the total potential field at the current time. Tangential guiding item and damping term The expected maneuver trend obtained by the combined action is used to indicate the direction of obstacle avoidance / collision avoidance maneuver, and its magnitude is used to characterize the maneuver intensity. As an intermediate quantity, it is not directly used as the final control command output, but rather through the projection operator. Mapped to a set of feasible controls that satisfy ship maneuvering constraints To obtain executable control variables .

[0250] In one embodiment, the feasible control set It consists of constraints on yaw rate and longitudinal acceleration. ,and , The projection operator To be The operator is mapped to the control quantity that satisfies the above constraints.

[0251] Will Projected onto feasible control set :

[0252] ;

[0253] When potential field convergence stagnation is detected, such as... Furthermore, if the risk does not decrease, a replanning or disturbance restart mechanism will be triggered.

[0254] Output: Stability control command.

[0255] Example 2:

[0256] Figure 2 This is a schematic diagram of the system used to implement a potential field-based ship navigation early warning method according to an embodiment of the present invention. Figure 2Each module is implemented by a computer program and is used to execute the corresponding steps in the method of the embodiments of the present invention. The specific steps executed by each module are described in the method embodiments above. Specifically, a ship navigation early warning system according to an embodiment of the present invention includes:

[0257] (1) Data acquisition and fusion module 10: It is used to receive inputs from radar, vision, inertial navigation, electronic chart, rule base and / or environmental sensing through ship status input interface 81, target ship status information input interface 82, channel / obstacle geometry input interface 83, rule base interface 84 and environment and uncertainty input interface 85 respectively, and output unified status and uncertainty;

[0258] (2) Scale unification and risk mapping module 20: used to perform dimensionless scale unification and risk mapping normalization on the information collected by the data acquisition and fusion module 10;

[0259] (3) Rule Triggering and Constraint Generation Module 30: Used for encounter situation recognition, rule triggering, conflict priority processing, and output. ;

[0260] (4) Potential field construction and fusion module 40: used for construction , , , and integration ;

[0261] (5) Risk warning module 50: used for calculation Warning level ( Contribution decomposition; and based on the costs or constraints comprised of potential field, rules, and uncertainties, in or warning level ( When the conditions for triggering collision avoidance decision optimization are met, the collision avoidance decision optimization module 60 is triggered;

[0262] (6) Collision avoidance decision optimization module 60: used to solve constraint optimization and output executability output such as manipulation instructions;

[0263] (7) Command output and interpretation module 70: outputs heading, speed and / or rudder angle commands to the ship execution and control system 91 such as autopilot and / or navigation controller, and outputs interpretation information to the driver or shore via human-machine interaction display terminal 92 and / or shore-based / supervisory communication interface 93.

[0264] In one specific implementation, the potential field construction and fusion module and the collision avoidance decision optimization module are updated on a rolling basis with a fixed control cycle, and the target ship's state and uncertainties are updated and the total potential field is reconstructed in each control cycle. Figure 6 A schematic diagram of the entire rolling-optimized collision avoidance closed-loop structure is shown. Figure 6The rolling optimization collision avoidance closed-loop control includes: 601, obtaining the current state and environmental input; 602, inputting the obtained information into the prediction model, including kinematic or dynamic models; 603, obtaining the cost structure including the total potential field; 604, obtaining the constraint structure including operational constraints and rule constraints C; 605, optimizing using an optimization solver such as SQP, iLQR, or gradient projection; 606, executing control based on the optimization output; 607, controlling the ship to execute corresponding commands and environmental evolution; 608, entering the next cycle of rolling feedback update; The optimizer solution process also includes: 609, determining whether there is an initial local optimum or oscillation, and if so, triggering the corresponding replanning; This step can suppress local optima and oscillations in the potential field method and maintain stability and robustness in scenarios with multiple ships densely packed and close to the edge in narrow waterways.

[0265] Example 3:

[0266] The present invention also provides an electronic device, the device including a processor, a memory, a bus, and a computer program stored in the memory and executable on the processor. The processor includes one or more processing cores, the memory is connected to the processor via the bus, and the memory is used to store program instructions. When the processor executes the computer program, it implements the steps in the above-described method embodiment of Embodiment 1 of the present invention.

[0267] Furthermore, as an executable solution, the electronic device can be a computer unit, which can be a desktop computer, laptop, handheld computer, cloud server, or other computing device. The computer unit may include, but is not limited to, a processor and memory. Those skilled in the art will understand that the above-described structure of the computer unit is merely an example and does not constitute a limitation on the computer unit. It may include more or fewer components, or combine certain components, or use different components. For example, the computer unit may also include input / output devices, network access devices, buses, etc., and this embodiment of the invention does not limit this.

[0268] Furthermore, as an executable solution, the processor can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or any conventional processor, etc. The processor is the control center of the computer unit, connecting various parts of the entire computer unit via various interfaces and lines.

[0269] The memory can be used to store the computer programs and / or modules. The processor implements various functions of the computer unit by running or executing the computer programs and / or modules stored in the memory and by calling data stored in the memory. The memory may mainly include a program storage area and a data storage area. The program storage area may store the operating system and at least one application program required for a function. The data storage area may store data such as ship status, target ship list, channel / obstacle geometry, rule entries, potential field parameters, risk index, and control commands. In addition, the memory may include high-speed random access memory and non-volatile memory, such as hard disk, RAM, plug-in hard disk, smart media card (SMC), secure digital (SD) card, flash card, at least one disk storage device, flash memory device, or other volatile solid-state storage device.

[0270] Example 4:

[0271] The present invention also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the steps of the method described in the embodiments of the present invention.

[0272] If the modules / units integrated in the computer unit are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the methods of the above embodiments can also be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include: any entity or device capable of carrying the computer program code, recording media, USB flash drives, portable hard drives, magnetic disks, optical disks, computer memory, read-only memory (ROM), random access memory (RAM), and software distribution media, etc. It should be noted that the content contained in the computer-readable medium can be appropriately increased or decreased according to the requirements of legislation and patent practice in the jurisdiction.

[0273] Example 5:

[0274] The present invention also provides a computer program product, including a computer program that, when executed by a processor, implements the steps described above.

[0275] Although the invention has been specifically shown and described in conjunction with preferred embodiments, those skilled in the art should understand that various changes in form and detail may be made to the invention without departing from the spirit and scope of the invention as defined in the appended claims, all of which shall be within the scope of protection of the invention.

Claims

1. A ship navigation early warning method based on potential field, characterized in that, include: S1, acquire the ship's status, target ship status, set of static obstacles, set of navigation rules, and environmental disturbance and uncertainty information; the ship's status and target ship status include the ship's ground-fixed coordinates, heading, and speed; wherein, static obstacles include one or more of the following: channel boundaries, shorelines, shoals, reefs, bridge piers, and harbor structures; the environmental disturbance and uncertainty information includes a set of estimated covariances or standard deviations obtained by statistically analyzing wind, wave, and current estimations, visibility, sensing errors, and / or execution errors; S2, using the selected length scale Speed ​​scale With time scale The ship's state, target ship state, static obstacle set, navigation rule set, and environmental disturbance and uncertainty information obtained in step S1 are processed to be dimensionless, resulting in dimensionless state quantities and dimensionless indices derived from these state quantities. The dimensionless indices include: dimensionless distance indices, dimensionless time indices, and dimensionless uncertainty scales. These dimensionless indices are then input into a risk mapping function. Perform normalization; Wherein, the risk mapping function for any dimensionless index s for: ； in, Here is the dimensionless reference value corresponding to the dimensionless index s. The gain coefficient is greater than 0; This represents a nonlinear saturation function or activation function; S3, based on trajectory points Shortest distance to static obstacles in the static obstacle set and preset safety margin The difference is used to determine the trajectory points using the risk mapping function. static risks And based on the static risk and the predetermined logarithmic potential energy transformation formula, the ship's trajectory point is obtained. Static environment potential field ; The predetermined logarithmic potential energy transformation formula is as follows: ; Among them, the The dimensionless shortest distance obtained under the length scale L determined in step S2, wherein The dimensionless safety margin obtained under the length scale L; in, ; ； in, For a predetermined minimum positive number, For the static environmental potential field of index s, For the shortest distance The static potential field of the environment; S4, based on the urgency index of the relative motion between the two ships and the preset interactive safety distance, obtains the distance between the current ship and each target ship. Interaction risks According to interaction risk The predetermined logarithmic potential energy transformation formula is used to obtain the relationship between the ship and each target ship. Dynamic interactive potential field The dynamic interactive potential field between the ship and all target ships is obtained through a predetermined multi-objective weighted fusion method. The urgency index includes the dimensionless nearest point distance between the two ships obtained using the length scale L. and the time of closest arrival of the dimensionless point ; S5, based on this ship and each target ship The encounter status recognition result triggers navigation rules, generating a target ship. rule constraint set ; and based on control quantity and the set of rules and constraints Constructing the permissible maneuver set and prohibited motor assembly To then construct the target ship Corresponding distance-based violation Then according to The target ship is obtained by using the predetermined logarithmic potential energy transformation formula. The regular potential field components The rule constraint potential field of this ship for all target ships is obtained by weighted fusion of the rule potential field components of all target ships. The control quantity Including yaw rate and longitudinal acceleration; where: ； in, This is the penalty coefficient; Indicates the target ship The corresponding urgency weight, The total number of target ships; express violation The depth, where, when hour, ;otherwise, ; Indicates control quantity With permitted maneuver sets The minimum Euclidean distance; S6, based on the aforementioned environmental disturbance and uncertainty information, for distance-related indicators... Make corrections to obtain the corrected effective distance. and use the aforementioned in distance-based risk mapping Replace the distance-type indicator To obtain an uncertainty-corrected potential field ;in: in, The distance index is the dimensionless index after step S2. Confidence coefficient; S7, the static environment potential field, the dynamic interaction potential field, the rule constraint potential field, and the uncertainty correction potential field are weighted and fused according to predetermined weights to obtain the total potential field. ; S8 represents the total potential field at time t. Mapping to a risk index via exponential saturation or the Sigmoid function. , And by comparing risk indices The warning level is output based on the value of the preset risk index threshold.

2. The ship navigation early warning method based on potential field according to claim 1, characterized in that, express Compared to the prohibited maneuver assembly Signed distance to the boundary: when When the value is positive and equal to Minimum distance to the boundary of the prohibited movement assembly; for any element in; when When the value is 0 or negative, it is used to make Output 0 if no violation occurs; Used to reflect the strength of the contribution of different target ships to the rule constraints; in, The definition of is: ； in, It is a Euclidean norm; In step S6, the Includes: shortest distance to static obstacles , and the target ship nearest point distance and channel boundary margin At least one of them; when When static obstacles are involved, the state estimation covariance shall include at least the ship's state covariance. ;when Involving the target ship At that time, the state estimation covariance includes at least the ship's state covariance. Covariance with the target ship's state and form relative covariance .

3. The ship navigation early warning method based on potential field according to claim 1, characterized in that, Step S8 further includes one or more of the following steps: Multiple risk index thresholds can be preset to output multiple warning levels; the higher the risk index threshold, the higher the warning level. Output the percentage of each potential field's contribution to the total duration; Using the potential field evolution as a risk index To provide early warning, at this time: The total potential field at time t ; , and These are the preset weighting coefficients.

4. The ship navigation early warning method based on potential field according to claim 1, characterized in that, The process after step S8 also includes: S9, when the risk index Greater than or equal to the preset collision avoidance trigger risk threshold or the rate of change of the risk index Greater than or equal to the preset collision avoidance trigger risk change rate threshold Collision avoidance optimization is triggered in a timely manner, and a multi-objective collision avoidance optimization problem is solved under preset ship kinematics or dynamics constraints, maneuvering constraints, and rule constraints to generate executable heading, speed, rudder angle, and / or yaw rate control commands, which are then output to the ship control system; wherein, the risk index change rate is defined as: ； To control the period or time step; the multi-objective collision avoidance optimization problem uses a control sequence. Let be the decision variables, where , Yaw angular velocity, For longitudinal acceleration, the prediction step size is... The predicted trajectory point position vector is , The reference point is the centerline of the navigation channel or the target trajectory; the objective function used in the multi-target collision avoidance optimization is: ； in, For the first The total potential value at the step prediction time. These are the weighting coefficients for the corresponding items. Let be the Euclidean norm, and let... , This is the control quantity that was output and executed in the previous control cycle.

5. The ship navigation early warning method based on potential field according to claim 1, characterized in that, In step S4: ； ； in, For this vessel and the target vessel The nearest distance between them; To predict the relationship between this vessel and the target vessel The closest time between; Define the time window function: ； Preset interaction safety distance for: ； ， ； ； ； ； ； ； in, and These are the lengths of the current ship and the target ship j, respectively. For the ship's speed, For the target ship speed, Indicates the target ship The relative position vector of the ship itself. Indicates the target ship The relative velocity vector with respect to the ship itself; This is the time constant for the ship's reaction or execution; Indicates the relationship between this vessel and the target vessel The dimensionless uncertainty scale of the relative position is obtained by propagating the covariance of the position estimates of the ship and the target ship, and is used to correct the interaction safety distance. , and These are preset weighting coefficients; For indicator functions, only when the following conditions are met The value of the indicator function is only non-zero when the specified time is reached; This is the upper limit of the time window; The decay time constant; This is a preset constant; The total number of target ships; and This is the preset gain coefficient; The preset time threshold; where, All are dimensionless indices obtained after dimensionless processing in step S2, and and To be expressed on the same length scale.

6. The ship navigation early warning method based on potential field according to claim 1, characterized in that, It also includes one or more of the following: In step S6, an uncertainty correction potential field is obtained by constructing an upper bound term for the risk index. ;in: ； in, The penalty weight for uncertainty has a value ≥ 0; Let be the trace of the covariance matrix; where, when only static barriers are considered, This is the covariance of the ship's state; when the target ship is involved... When the relative amount is, For this vessel and the target vessel The relative covariance, and under independent estimation conditions ; In step S3, if a channel boundary exists, then the signed distance is used as the basis. Constructing the boundary potential field To strengthen penalties for crossing the boundary, and Merging Among them, when the trajectory point When within the boundary, When the trajectory point When at the boundary, When the trajectory point When crossing the boundary, .

7. The ship navigation early warning method based on potential field according to claim 4, characterized in that, Also includes: The multi-objective collision avoidance optimization problem is solved using SQP, iLQR, gradient projection, or rolling MPC methods to obtain the control sequence. Only the first item is executed. And will be updated on a rolling basis in the next cycle; When detected If the risk does not decrease or the control quantity frequently changes direction, trigger tangential guidance, damping term, random disturbance restart or replanning strategy, and project the generated guidance quantity onto the feasible control set. ; in, Represents the total potential field gradient, express The Euclidean norm; It is a pre-defined small positive number.

8. An electronic device, characterized in that, It includes a memory and a processor, the memory storing at least one program, the at least one program being executed by the processor to implement the steps of the potential field-based ship navigation early warning method as described in any one of claims 1 to 7.

9. A computer-readable storage medium, characterized in that, The storage medium stores at least one program segment, which is executed by a processor to implement the steps of the potential field-based ship navigation early warning method as described in any one of claims 1 to 7.

10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by the processor, it implements the steps of the potential field-based ship navigation early warning method as described in any one of claims 1 to 7.

Images